Epigenomic tensor predicts disease subtypes and reveals constrained tumor evolution.

Area of Science:

• Computational biology
• Epigenomics
• Bioinformatics

Background:

• Understanding disease-specific epigenomic alterations is crucial for diagnosis and treatment.
• Complex patient data presents challenges in identifying robust epigenomic differences.

Purpose of the Study:

• To present DeCET (decomposition and classification of epigenomic tensors), an integrative computational approach.
• To identify epigenomic differences among tissue types, differentiation states, and disease subtypes.

Main Methods:

• DeCET simultaneously analyzes hierarchical heterogeneous epigenomic data.
• Applied to patient data from uterine benign tumors (leiomyoma) and public epigenomic datasets.

Main Results:

• Identified distinct epigenomic features discriminating normal myometrium and leiomyoma subtypes.
• Leiomyomas show alterations in distal enhancers and long-range histone modifications within chromatin contact domains.
• Demonstrated DeCET's effectiveness on diverse cancer and cellular state epigenomic datasets.

Conclusions:

• DeCET facilitates the identification of robust epigenomic differences in complex datasets.
• Epigenomic features extracted by DeCET enhance understanding of disease states, development, and differentiation.
• This approach supports future therapeutic, diagnostic, and prognostic strategies.